00:00
<gsnedders>
gavinc: s#and#and/or# :)
00:16
<gavinc>
gsnedders: good point.
02:17
<Hixie>
hmm
02:17
<Hixie>
how the hell do i serialise a method pointer in JS
02:17
<Hixie>
bummer
05:58
<Hixie>
annevk: was it taken out again? i thought .append() and co were in, that's why i used them in the examples
05:58
<annevk>
those are
05:58
<annevk>
but createElement with multiple arguments is not
05:58
<Hixie>
ah, ok
06:03
<annevk>
Lachy: http://dvcs.w3.org/hg/domcore/raw-file/tip/Overview.html#errors now uses "name" rather than "type" throughout (IDL has been updated too)
06:07
<hsivonen>
UA string evangelism makes me sad
06:08
<hsivonen>
Mozilla added "Mobile" and "Tablet" tokens to give sites something other than "Android" to sniff for. Now we are debating whether B2G has to say "Android" anyway, because the string with Android got evangelized.
06:09
<annevk>
hmm http://lists.w3.org/Archives/Public/public-webrtc/2012Jun/ pretty much no involvement from people with some history with API design...
06:09
<annevk>
does explain http://lists.w3.org/Archives/Public/public-script-coord/2012AprJun/0269.html ...
06:09
<heycam>
I'm not in the mood to read those last 100 mails, tbh
06:09
<annevk>
me neither
06:10
<annevk>
maybe I'll read something when they get around to publishing again
06:10
<heycam>
yeah
06:10
<heycam>
I must say, the spec seems pretty unreadable to me
06:10
<heycam>
the formatting
06:10
<annevk>
it's a mixture of WHATWG HTML plus whatever the new guy is using
06:10
<annevk>
plus probably some other things
06:12
<heycam>
I wonder if we should mark stringifiers as not for use unless you really have to
06:12
<heycam>
to me they seem like they invite author confusion
06:16
<hsivonen>
heycam: SVG scripts can nest like this <svg><script>foo();<script>bar();</script>baz();</script> thanks to XML-likeness
06:16
<heycam>
hsivonen, you can do that with HTML too with DOM calls right
06:16
<hsivonen>
heycam: yes, but then the script wont be parser-inserted
06:17
<shepazu>
couldn't we simply say that <script> can't nest?
06:17
<heycam>
hsivonen, is the behaviour for parser inserted nested script elements undefined?
06:18
<hsivonen>
shepazu: that would be useless without making it impossible for them to nest
06:18
<heycam>
hsivonen, (beacuse html can't have them?)
06:18
<hsivonen>
heycam: no
06:18
<shepazu>
that <script> is not a valid child of <script>, and that any content doesn't execute?
06:18
<hsivonen>
shepazu, heycam: what I said about the added parser complexity was an aside. Please don't try to fix it by making it worse.
06:19
<heycam>
lol ok, I had hoped "do the same as HTML" should be making things better
06:19
<shepazu>
I'm not convinced that this is a real-world problem
06:19
<hsivonen>
All I need to do is to put the "had async of defer" bit on the stack when now one bit on the tree builder is enough
06:19
<heycam>
ok, so it wasn't something to be avoided then
06:20
<hsivonen>
heycam: right
06:20
<heycam>
btw, what is the behaviour of non-parser inserted nested script elements in html?
06:20
<hsivonen>
heycam: undefined, which is a spec bug
06:20
<heycam>
aha
06:20
<hsivonen>
(browsers disagree, too)
06:20
<hsivonen>
(though IE10 and WebKit agree)
06:20
<heycam>
the best kind of spec bug
06:21
<hsivonen>
but parser-inserted nested script are defined. Yay.
06:21
<Hixie>
undefined?
06:21
<heycam>
heh, funny
06:21
<Hixie>
in which spec?
06:22
<hsivonen>
Hixie: HTML Living Standard
06:22
<hsivonen>
Hixie: well, at least at least 3 channel regulars failed to find the definition
06:22
<Hixie>
hsivonen: what's undefined? i thought the script processing was pretty well-defined by this point
06:22
<heycam>
not really knowing how any of this works currently, I would just say the textContent of the outer script element is the script that's run
06:22
<hsivonen>
Hixie: so it's undefined in practice even if you could find the definition
06:23
<hsivonen>
heycam: and you'd be wrong :-)
06:23
<Hixie>
hsivonen: sure, but the differnce is that in one case i have to figure out how to fix it to actually define it, and in the other i just have to add a note
06:23
<Hixie>
hsivonen: so it's an important difference to me :-)
06:23
<heycam>
hsivonen, not surprised, as it's just something sensible :)
06:23
<hsivonen>
Hixie: we need two things:
06:23
<hsivonen>
1) something that says that upon inserting a DocumentFragment, the whole fragment gets inserted before attempting to execute any of its scripts
06:23
<Hixie>
(btw, personally i think the way svg-in-html script parsing is defined now is fantastic, as it is simpler than html, and that can imho only be a good thing)
06:24
<Hixie>
hsivonen: hm, i can believe that behaviour of script insertion with document fragments isn't well-defined
06:24
<hsivonen>
2) a definition for the order in which the scripts that got inserted are attempted to execute
06:24
<Hixie>
hsivonen: since the html text predates dom core being written
06:24
<Hixie>
hsivonen: k. is there a bug/e-mail on this?
06:25
<hsivonen>
Hixie: there's a bug about #1 and it has an off-topic comment about #2
06:26
<Hixie>
hsivonen: k
06:26
<jgraham>
heycam: (textContent would be very confusing <script>alert(1); <script>alert(2)</script></script> would alert 2 then 1 then 2
06:26
<jgraham>
)
06:27
<jgraham>
(happily that isn't what happens)
06:27
<heycam>
jgraham, forgot to also add: the nested <script> elements would not get executed themselves
06:27
<hsivonen>
Hixie: oh and for nested scripts, both points need to apply also when inserting a tree of elements without a document fragement wrapper
06:27
<hsivonen>
jgraham: not even in Opera?
06:27
<Hixie>
same bug #?
06:28
<Hixie>
17543?
06:28
<jgraham>
heycam: It would be very confusing not to run some scripts depending on their ancestors
06:29
<jgraham>
hsivonen: I'm not sure what happens in Opera. iirc *in SVG* that case works OK, but there might be that bug if you nest scripts in HTML
06:29
<heycam>
(disagree, especially if the rule is "if there's an ancestor <script> it doesn't run")
06:30
<heycam>
(sounds like it's moot anyway tho)
06:30
<jgraham>
heycam: Pretty much, everyone agrees on that much behaviour at least
06:30
<annevk>
Hixie: the plan is to add hooks to "DOM" for "elements inserted" and "elements removed" which can be used by HTML to then do stuff
06:30
<Hixie>
and 17529 i guess
06:30
<Hixie>
annevk: k
06:30
<annevk>
Hixie: they're both plural because the idea is that you can atomically insert and remove more than one element
06:31
<heycam>
annevk, sorry for hammering on about the event handler stuff again, my small brain forgets previous conversations about things like that :)
06:31
<jgraham>
hsivonen: Doesn't XHTML give you parser-inserted nested scripts?
06:31
<annevk>
heycam: no worries
06:31
<hsivonen>
jgraham: yes
06:31
<hsivonen>
Hixie: https://www.w3.org/Bugs/Public/show_bug.cgi?id=17529
06:32
<jgraham>
I guess that is another case to write tests for
06:33
<jgraham>
The platform is way too complicated. Every time you think "I should test this simple thing" you realise there are multiple contexts in which you need to test it, often with different behaviour
06:35
<jgraham>
Anyway, time to cycle to the office (in the sunshine!), I guess
06:35
<annevk>
Sweden summertime ftw
06:36
<Hixie>
i cycle to work every day i go to work... California time ftw :-P
06:48
<AryehGregor>
I now live in a place where it's too hot to bike in the summer, too rainy to bike in the winter, and too hilly to bike ever. \o/
06:48
<AryehGregor>
(doesn't bother me, I don't bike anyway)
07:09
<jgraham>
annevk: It's OK it's forecast to rain next week
07:10
<annevk>
aaah
07:10
<jgraham>
so you will miss the authentic Swedish summer experience
07:10
<annevk>
I don't feel too bad about that I think
08:13
<Ms2ger>
annevk, once I figure out what you guys need me to do :)
08:13
<annevk>
Ms2ger: so in principle you're okay with merging and all?
08:14
<annevk>
Ms2ger: I guess maybe lack of input should be considered as "okay" but this is not a mailing list :p
08:14
<Ms2ger>
Yes, of course
08:14
<annevk>
kk
08:14
<annevk>
I have no insight in the line of thinking of robots unfortunately
08:15
<Ms2ger>
:D
08:25
<odinho>
21:43 < Ms2ger> annevk, you're wrong :)
08:25
<odinho>
Where is the context for all of this? :S
08:25
<odinho>
If anywhere at all :]
08:25
<Ms2ger>
Private email
08:25
<odinho>
Ah, okay, so not then.
08:26
<Ms2ger>
He said not-nice things about Mercurial
08:26
<odinho>
Ms2ger: I can do that too. :]
08:26
Ms2ger
curses Operians
08:27
<annevk>
the context is a discussion with berjon about merging ReSpec and Anolis bibref and xref databases
08:31
jgraham
has spent quite a lot of the morning trying to help various people with git. But at least helping people with git doesn't start "learn mq".
08:32
<annevk>
Ms2ger: can you review the commit I just made?
08:32
<annevk>
Ms2ger: don't want Hixie to run into issues when he starts fixing the HTML side
08:33
<annevk>
jgraham: :)
08:33
annevk
is promoting git in the absence of any experience with it
08:33
<annevk>
I guess at some point we should move the specs over
08:33
<annevk>
to github
08:33
<annevk>
so everyone can do pull requests and other fancy things
08:34
<Ms2ger>
Whoa
08:34
<Ms2ger>
Error names
08:35
<Ms2ger>
annevk, s#and/or#or#
08:35
<annevk>
really?
08:35
<annevk>
because it's "and" for the replace stuff
08:35
<Ms2ger>
and/or is an abomination in the eyes of the Lord
08:36
<odinho>
Git was my first VCS. Right around the time Linus unveiled his first plumbing layer Git, I was in the market for a VCS. So I took it, warts and all. Used cogito a bit, but that was eventually merged into Git - and today it's really nice and helpful. :-)
08:36
Ms2ger
doesn't like it much either
08:36
<annevk>
Ms2ger: which Lord though?
08:36
<odinho>
Ms2ger: Lord Voldemort?
08:36
<Ms2ger>
odinho++
08:36
<annevk>
odinho: dude, cannot mention his name
08:36
<Ms2ger>
odinho, also, that's called "Stockholm syndrome"
08:36
<odinho>
annevk: fsck, didn't remember.
08:38
<annevk>
oh well, fixed
08:39
<Ms2ger>
Thanks :)
08:47
<annevk>
ooh, no logs
08:54
<Ms2ger>
http://logbot.glob.com.au/ has logs
09:24
<annevk>
Fullscreen thread is full of talk
09:25
<MikeSmith>
an no walk?
09:26
<annevk>
right, except for fantasai, although the condescending "in the CSS WG we do..." she could have left out I think
09:27
<MikeSmith>
hahaha
09:28
<MikeSmith>
"in NN WG, we have a batshit insane way of doing many things"
09:28
<MikeSmith>
just FYI
09:40
<matjas>
I’d like to get a few lines of CSS added to http://www.whatwg.org/style/specification — who do I have to ask?
09:41
<matjas>
basically, this is all it takes to make a spec Responsive™: http://pastebin.com/raw.php?i=zNJV7VSF
09:41
<matjas>
for an example, try http://mathias.html5.org/specs/web-ecmascript/ on a small viewport
09:43
<odinho>
Hmm. Should make usercss easy to have on mobile phones. I'd like to tweak quite a few sites to have some responsive stuff thrown in.
09:44
<zcorpan>
matjas: wouldn't that break with the margin boxes in the html spec?
09:45
<matjas>
zcorpan: possibly, but i’m sure it can be done without any breakage
09:47
<zcorpan>
matjas: why have different padding above/below the breakpoint?
09:47
<matjas>
zcorpan: try it, it looks better / is more readable on small screens
09:48
<matjas>
(IMHO)
09:48
<zcorpan>
it looks bad when width is 60em+1px :-P
09:49
<zcorpan>
anyway, i guess Hixie is the one to ask
09:50
<zcorpan>
though it's not unusual for specs to just have local overrides
09:57
<matjas>
Hixie: please responsify the spec-reading experience in the mobile workspace, increasing the end users’s social mindshare by leveraging Web 3.0 CSS5 Media Queries
09:58
<odinho>
matjas: WAT, don't speak like that.
10:00
<matjas>
odinho: sorry, i got caught up in the moment
10:01
<odinho>
matjas: A big corporate faceless PR/hype/buzzword-over-substance moment?
10:01
<matjas>
I can’t believe I forgot to mention SEO :'(
10:20
<annevk>
matjas: probably best to email or file a bug
10:21
<annevk>
matjas: canAskHixieOnIRC = Hixie.around() && !Hixie.busy()
10:21
<matjas>
hah
10:24
<annevk>
what is the obj.is(Node) discussion about? And why is it only a problem for DOM?
10:24
<annevk>
I'm missing something...
10:25
<Ms2ger>
Cross-window instanceof?
10:29
<annevk>
why can't we make instanceof cross-window?
10:29
<Ms2ger>
Dunno
10:29
<annevk>
also
10:30
<annevk>
then this isn't just the dom, but also applies to XMLHttpRequest, Worker, EventSource, etc. no?
10:31
<Ms2ger>
Presumably obj.is(XMLHttpRequest) would be expected to work too
10:46
<jgraham>
That's why people preferred Node.isInterface(obj) or whatever it was
10:48
<jgraham>
But it does seem a bit silly to have to implement it on all objects rather than having an operator
10:48
<jgraham>
Well all interface objects
10:49
<annevk>
not sure why an interface object should be different from other javascript objects really
10:49
<annevk>
interface is just a convenience when writing specs
10:50
<jgraham>
Right, you would expect otherWindow.Object.isInterface({}) to work as well, I guess
11:00
<annevk>
http://annevankesteren.nl/2012/06/notifications
12:26
<annevk>
jgraham: chaals' argument for iconAlt is somewhat compelling actually
12:27
<charl>
annevk: interesting post but quite honestly i don't get this part: perm == "granted" ? true : false
12:27
<charl>
so you are basically saying "if true then true, if false then false"
12:28
<charl>
you test against a boolean to produce exactly the same boolean
12:28
<annevk>
I will drop the ? true : false and then go into hiding :)
12:28
<charl>
lol
12:29
<annevk>
thanks
12:29
<charl>
np!
12:30
<jgraham>
annevk: I guess
12:31
<annevk>
I wish hober, jonlee|afk, and John would chime in
12:31
<annevk>
but sleep
12:34
<AryehGregor>
Hmm. Were my reflection tests ever actually approved?
12:34
<AryehGregor>
I thought they were supposed to be at some point by default.
12:35
<AryehGregor>
But they're wrong by now, because Hixie keeps changing the spec. :(
12:36
jgraham
is sort of ignoreing the whole approved/unapproved thing
12:36
<annevk>
it was a bullshit concept to begin with
12:37
<AryehGregor>
Well, I wouldn't go that far.
12:37
<annevk>
it helps forcing review somewhat of badly written tests
12:38
<annevk>
other than that I don't really see the point of moving tests around and making them hard to find
12:41
<AryehGregor>
We want a "your browser passes X% of tests" kind of thing, for marketing purposes, and don't want Joe Q. Random (or, more likely, a browser vendor) to submit garbage tests to distort it.
12:41
<annevk>
we all know it's Microsoft
12:42
<AryehGregor>
Well, yeah.
12:43
<AryehGregor>
So: I guess I should write a rant yelling at public-css-testsuite for imposing stupid and onerous requirements on test submitters only after I've submitted all my tests, not before.
12:44
<AryehGregor>
Preferably once they're approved.
12:44
<jgraham>
It doesn't really stop that
12:44
<jgraham>
See also the CSS 2.1 outline tests (or something)
12:44
<AryehGregor>
No, but it means that if someone notices, you can revoke the approval at that point.
12:45
<jgraham>
The tests I'm thinking of were all the same but for different colours
12:45
<annevk>
also written by Microsoft
12:45
<jgraham>
would have red outlines work, but blue ones fail?
12:46
<Ms2ger>
Maybe!
12:46
<AryehGregor>
And did they happen to test a trivial feature that IE supported but no one else did?
12:46
<annevk>
philosoraptor having a field day
12:46
<AryehGregor>
I'm all in favor of testing every possible test color, but only if it's all in one test file.
12:46
<AryehGregor>
2^24 tests isn't a lot by the standards of today's CPUs, right?
12:46
<AryehGregor>
Hmm, but if you add alpha . . .
12:47
<jgraham>
Well Opera doesn't support it. Which is a bug, but not 200 tests worth of bug (compared to the rest of the CSS testsuite)
12:47
<AryehGregor>
Isn't the CSS 2.1 test suite actually pretty big?
12:47
<jgraham>
Not all that big
12:48
<Ms2ger>
9000 tests or so?
12:49
<AryehGregor>
Oh, that's nothing.
12:49
<AryehGregor>
I wrote a few hundred reftests just for transforms.
12:49
<AryehGregor>
And, of course, a couple of JS files that test like a billion things.
12:49
<Ms2ger>
Compared to the 650 000 tests I imported into m-c this morning...
12:50
<smaug____>
number of tests doesn't mean much
12:50
<smaug____>
some tests may test several things
12:50
<smaug____>
and many tests may test the same thing
12:50
<AryehGregor>
Ms2ger, that's only a few files, though.
12:50
<AryehGregor>
Granted, thousands of LOC.
12:51
<Ms2ger>
True
12:51
<smaug____>
hsivonen: new patch uploaded
12:51
<Ms2ger>
Two Mozillians chatting doesn't mean this is #developers, smaug____ :)
12:52
<AryehGregor>
Ms2ger, see, I'm not the only one!
12:52
<smaug____>
Ms2ger: oop, wrong tab :)
12:52
<smaug____>
oops
13:11
<odinho>
You keep doing that :P We just got ice cream out in the sun by the lake here. It was A+.
13:12
Ms2ger
kicks odinho
13:20
<annevk>
aah, we had ice-cream in the canteen
13:21
<annevk>
technically I guess it was the living room area on the other side
13:22
<jgraham>
It's all one room
13:22
<jgraham>
We aren't pretentious enough to subdivide
13:24
<annevk>
just enough to correct me? :p
13:24
<odinho>
annevk: So better at HQ then. :] Was quite pictoresque.
13:26
<jgraham>
But we have nicer chairs :p
13:26
<Ms2ger>
You should donate them to the HTMLWG
13:27
<annevk>
:)
13:59
<hsivonen>
Is Glenn Adams an Editorial Assistant now? https://www.w3.org/Bugs/Public/show_bug.cgi?id=17563
14:01
<Ms2ger>
Maybe he could start with editing the spec he's supposed to edit
14:01
<hsivonen>
(of course, trying to ban the use of canvas for decoration would be an exercise in futility, so I don't disagree with dismissing the bug)
14:02
<annevk>
hmm URLs
14:14
<annevk>
so is it better to have a tokenizer like this http://wiki.whatwg.org/wiki/URL#Parsing
14:16
<annevk>
or should the language be "split X on the first ":" if any and let (A, B) be the result" "if A starts with a character in the range RR and the remaining characters are in the range OO set scheme to A and X to B" "otherwise relative steps here..."
14:17
<annevk>
that is more how the spec is now
14:39
<annevk>
poor hober http://w3cmemes.tumblr.com/post/25563333689/bad-luck-brian-thought-html-wg-meetings-were-bad
14:47
<MikeSmith>
heh
14:48
<odinho>
:/
15:54
<dglazkov>
good morning, Whatwg!
15:55
<Ms2ger>
Morning
16:03
<MikeSmith>
good 01:05am, crime partners!
16:04
MikeSmith
is enjoying participation in the usual weekly round of back-to-back Thursday night telcons
16:05
<MikeSmith>
things could be worse; at least my boss didn't make me attend the Tracking Protection WG f2f
16:08
<odinho>
:P
16:21
<hober>
MikeSmith: very funny :)
16:22
<Ms2ger>
Poor hober :)
16:22
<MikeSmith>
I was just speaking hypothetically!